Design And Implementation Of Internet Congestion Control Algorithms

With the development of the computer network, the demand for the quality of the service of network including the large bandwidth of network, the low delay and the low vibration is growing. The network congestion is a great factor affecting the network quality of service. Avoiding congestion and enhancing the congestion control to ensure the quality of service have become the focus of the network research. By evaluating the network state and predicting the incipient congestion, AQM (Active Queue Management) could drop packets purposefully, thus sources can be well informed the network state and then adjust its sending-rate. The RED algorithm is recommended by IETF as the only candidate algorithm, it is not perfect in terms of response time, stability, the adaptability of dynamic network environment and parameters setting.Due to the RED algorithm is sensitive to parameters setting, an improved RED algorithm combines with queue length and sending-rate inspection was proposed, this algorithm modified the maximum dropping probability according to the sending-rate of the source node and the bandwidth of the link. Thus, it improved the network congestion phenomenon. This algorithm was verified in NS2 network simulator. The result showed that the improved RED could adapt the change of network traffic effectively, keep the queue length steady, reduce the phenomenon of the queue overflow or the idle queue greatly.Due to the paroxysmal business or the vibrated flows, the throughput is not satisfied, a fuzzy control RED algorithm based on average queue length and the variance of average queue length was proposed in this thesis, this algorithm no longer set the fixed threshold for each queue, but inferred the packets dropping probability dynamically according to the network traffic. Simulations showed that this algorithm could improve the throughput performance of the TCP traffic. When the traffic vibrated acutely, it could make the TCP traffic become gentle in a short time and provide the ideal performance improvement for Multimedia applications.